Arm-carried unmanned aerial vehicle near-earth target grabbing planning method based on eagle-tenon hunting arm swinging behavior

Abstract

The invention discloses an armed unmanned aerial vehicle near-earth target grabbing planning method based on an eagle-tenon hunting swing arm behavior, and the method comprises the steps: 1, carrying out the eagle-tenon hunting analysis and armed unmanned aerial vehicle system modeling, dividing a grabbing process into three stages, i.e., a pre-grabbing stage, a middle grabbing stage and a post-grabbing stage, and building an unmanned aerial vehicle motion constraint according to the eagle-tenon behavior characteristics of each stage; 2, designing a grabbing planning optimization problem taking energy optimization as a target, and introducing a position relaxation factor to adjust an arrival constraint; step 3, establishing planning constraints of each stage of eagle tenon approaching, grabbing and homeward flight swing arm behaviors; 4, dynamic planning constraints of the unmanned aerial vehicle system and the mechanical arm system are constructed; 5, solving a grabbing optimization problem and generating a track; and 6, outputting a grabbing planning track conforming to an eagle and tenon hunting arm swinging behavior. According to the method, through the design of the energy optimization target and the relaxation factor, the solving feasibility and real-time performance are enhanced while the physical constraint of the system is met, and good engineering realizability is achieved.

Description

Technical Field

[0001] This invention is a near-ground target grasping planning method for a drone with an arm based on the hunting arm swinging behavior of a falcon, belonging to the field of drone grasping design and mission planning. Background Technology

[0002] With the rapid development of intelligent and autonomous unmanned technologies, low-altitude economics has become a key research focus. Unmanned aerial vehicles (UAVs) have been widely used in various mission scenarios, including military reconnaissance, environmental monitoring, and facility inspection. When performing these typical tasks, UAVs usually select near-ground or ground targets as their targets. Mission execution is achieved through a process of rapidly and safely descending from a high-altitude cruising area to the near-target airspace, completing the specific task, and then climbing back to high altitude to continue flying. As a relatively new type of UAV, the robotic arm-equipped UAV, with its rich maneuverability and wide adaptability, has become a research hotspot in the UAV field. However, its maneuverability requires consideration of the collaborative action between the robotic arm and the UAV, resulting in inherent constraints on flight control. Therefore, when performing tasks requiring close proximity to the ground, specialized guidance algorithms must be designed to ensure the safe and efficient completion of the entire mission.

[0003] Currently, research on guidance algorithms for fixed-wing UAVs mainly focuses on cooperative control or tracking and grasping specific targets. This includes designing formation guidance methods that generate speed and heading commands, as well as research combining vector field guidance and model predictive control to address issues such as aerial obstacle avoidance and ground target tracking. These studies are mostly geared towards general guidance scenarios, lacking specific design for the "near-ground target approach" and "return" phases, which have different guidance target and state characteristics, resulting in limited effectiveness in near-ground guidance tasks. Furthermore, existing guidance laws are mostly based on custom rule designs, failing to systematically analyze and design from the perspective of system limitations and energy optimization. Therefore, there is an urgent need to design a guidance algorithm specifically for fixed-wing UAVs performing near-ground target tasks, which is crucial for expanding the effective application of fixed-wing UAVs in multiple fields.

[0004] In nature, birds of prey such as eagles and falcons exhibit remarkable maneuverability and environmental adaptability during hunting. They can flexibly adjust their dive attack trajectory according to the dynamics of their prey, achieving high-speed approach, precise guidance, and dynamic obstacle avoidance in a near-energy-optimal manner. Research shows that eagles employ a "delayed guidance" strategy when approaching their targets, with their turning adjustment exhibiting a process of gradual acceleration rather than a direct charge. Studies of species such as the Harris Eagle further reveal their cooperative hunting mechanisms, including a series of behavioral patterns of coordinated search, localization, and attack. These mechanisms have become important biomimetic sources for heuristic optimization algorithms. The entire process of an eagle swooping down from high altitude to capture a ground target and returning to the air is highly similar to the process of a drone performing a near-ground mission and returning to base. Establishing a mapping between the arm-swinging behavior of eagles during hunting and the near-ground target capture task of a drone with an arm can provide theoretical support and interpretability for the efficient execution of near-ground missions by drones.

[0005] In summary, this invention proposes a near-ground target grasping planning method for a drone with an arm based on the arm-swinging behavior of a falcon hunting. By analyzing the model of the falcon hunting process, action constraint relationships are established, and the falcon hunting arm-swinging behavior model is mapped to the "approach-grab-return" process of the drone with an arm performing near-ground target grasping. The planning method conforms to dynamic constraints, is simple and efficient, and the generated results have good executability and have certain practical application value. Summary of the Invention

[0006] The purpose of this invention is to provide a near-ground target acquisition planning method for UAVs with arms based on the arm-swinging behavior of falcons during hunting, aiming to solve the task planning problem of the entire near-ground target acquisition process for UAVs with arms. By drawing on the hunting process of falcons in nature, a model and constraints of the falcon's arm-swinging action for acquisition are established. Combined with the actual flight constraints of UAVs, a near-ground target acquisition planning problem for UAVs with arms is designed. By solving the planning problem, flight and acquisition commands are obtained, thereby realizing the acquisition of near-ground targets and the return flight after acquisition.

[0007] This invention proposes a near-ground target acquisition planning method for a drone with an arm based on the hunting arm swing behavior of a falcon. The specific implementation steps are as follows: Step 1: Falcon Hunting Analysis and Modeling of Arm-Mounted Unmanned Aerial Vehicle System S11. Analysis of the arm-swinging behavior of falcons during hunting. This invention draws inspiration from the specific hunting movements of falcons, focusing on the swinging motion of their legs and talons. Using existing experimental data and image materials, it analyzes that during the approach to prey, the falcon undergoes a delayed side-flight phase, gradually converging towards the prey. During this process, the talons swing forward to align with the prey. Specifically, to increase the lethality of the prey and reduce drag while carrying it, this grasping direction generally includes a smaller vertical component, primarily relying on horizontal velocity. This velocity direction is also related to the horizontal direction of the final return position, thus achieving a smooth grasping trajectory. Furthermore, since the falcon returns to high altitude while carrying prey, its legs generally swing backward during this process to provide lift and forward acceleration for the prey. This patent considers the grasping process in conjunction with the falcon's hunting arm swing, dividing it into three stages: pre-grab, during-grab, and post-grab. Based on the falcon's behavioral characteristics at each stage, optimization constraints are established to conform to the motion model of a drone with an arm. Finally, based on energy design optimization, trajectory generation is achieved.

[0008] S12. Establish a basic model of the boom-mounted drone. The robotic arm consists of a basic quadcopter drone and a robotic arm mounted below its center of gravity. The robotic arm is a two-jointed arm, with the first segment connected to the drone being [length missing]. The length of the second robotic arm is l 2 The robotic arms rotate around the y-axis of the drone's body coordinate system, and the system status includes the drone's three-dimensional position. 3D pose and the swing angle of the robotic arm Since the UAV has a self-stabilizing function, its attitude and position are considered to be coupled. Therefore, only the position task command needs to be designed in the planning part, and the UAV velocity variable is defined as follows. The angular velocity variable of the robotic arm is Based on the system design, the end-effector position and velocity can be calculated as follows: (1) The system will update its state by using discrete velocities with a fixed step size. The planning task requires generating the end-effector trajectory of the UAV with an arm, including position and velocity variables, and ensuring the capture of the target and its return with the target.

[0009] S13. Stage Division of Arm-Mounted Drone Grabbing Task Following the three stages of a falcon's hunting arm swing, the arm-mounted drone grasping task is divided into three stages: the arm-mounted drone approach stage, the target grasping stage, and the grasping and carrying flight stage, designated as Stage 0, Stage 1, and Stage 2, respectively. During optimization, the execution time for each stage is set as follows: The transition from Phase 0 to Phase 1 is marked by the arrival of the transition time. Once the end effector of the robotic arm of the drone reaches the position of the target to be grasped, the grasping preparation phase is complete. The transition from phase 1 to phase 2 is marked by the arrival of the transition time. Furthermore, the robotic arm of the drone can swing backwards according to the hunting arm swing characteristics of a falcon, achieving falcon-like arm swing angle constraints. Phase 2 completion marker, execution time reached. Furthermore, the end effector of the drone with the arm reached the return-to-home position.

[0010] Step 2: Optimize the near-ground target acquisition planning of the UAV with arm. This invention addresses the guidance and planning problem of drone and robotic arm motion during near-ground target grasping by a drone with an arm. To further map and correspond this to the hunting arm swinging behavior of a falcon, the guidance and planning problem is transformed into a planning optimization problem. The aim is to establish an optimization objective through energy relationships, while ensuring that the inherent physical constraints of the drone and the behavioral characteristics of the falcon's arm swinging behavior are satisfied before, during, and after grasping. Therefore, the state variables of the optimized system are determined as follows: The control input is .in, Location of the drone. The angle between the two joints of the robotic arm. For drone speed, Let be the angular velocities of the two joints of the robotic arm. Accelerate the drone Let be the angular acceleration of the two joints of the robotic arm.

[0011] S21. Design an optimization problem for near-ground target capture. The design optimization problem for a drone with an arm for grasping is as follows: (2) in, The cost function value, Total execution time Unmanned aerial vehicle system motion vector, , The angular velocity of the robotic arm. , , , All are relaxation factors. denoted as the relaxation factor proportional coefficient, the relaxation factor is used to constrain the reach-to-target constraint of the boom-armed UAV. This formula, through full-process integration, yields the comprehensive cost of the boom-armed UAV's near-ground target capture and return process, serving as the optimization objective for the planning task. The optimization problem requires minimizing the cost function by adjusting the input, thereby generating the desired system state variables for the entire process.

[0012] S22, Location Relaxation Factor Design S13 specifies that stage switching must satisfy both time and location arrival conditions. Since setting strict location and time conditions in optimization problems can easily lead to non-convexity and solution failure, a relaxation factor is designed in the specific optimization arrival design to reduce the constraints of the arrival conditions. Firstly, the target position is set to be captured upon completion of stage 0. When Phase 2 is completed, the end position of the robotic arm of the drone will be... Based on this, the relaxation factor is defined as: (3) in, The position vector of the robotic arm-mounted UAV can be adjusted by setting a relaxation factor to relax the constraints on the arrival position. In actual mission execution, since the end of the robotic arm is generally connected to a long mechanical gripper, it can still grasp the target even when there is a small error between the gripper and the object. This design meets the conditions for practical applications.

[0013] Step 3: Design planning constraints for mimicking the hunting arm swing behavior of eagles. The three stages of the falcon's hunting arm swing process correspond to the three stages of the drone's grasping action. Therefore, in designing the constraints of the optimization problem, optimization problems were designed according to different stages, as follows: S31, Stage 0: Planning Constraints for Arm-Mounted Unmanned Aerial Vehicles Approaching a Falcon-like Swing Arm (S31, Stage 0) Considering that eagles typically transition from a dive to a smooth, lateral grasping motion before approaching their prey, and pre-aime through arm swinging, at the end of phase 0, the end-effector velocity of the drone with the arm must not have any vertical velocity, and the robotic arm itself must be oriented horizontally. Therefore, the motion constraints are as follows: (4) in, To align the arm swing with the desired angle in a falcon-like hunting motion, it is set to... , , , The positive error boundary should be determined based on the actual situation and the solver performance. , Set the target speed to 0. As an angle relaxation factor, the UAV with arm is required to maintain the same speed as the target at the completion of phase 0 and to align itself according to the arm swing characteristics of the falcon hunting preparation phase. In addition, the constraint of this phase also includes the first formula of formula (3), that is .

[0014] S32, Phase 1 Planning Constraints for Arm-Modified UAV with Falcon-Inspired Grasping Arm During the capture of prey, falcons transition from aiming and grasping to loaded flight by swinging their arms backward, ensuring more efficient flight with prey. Considering this process, the present invention designs the following constraints for the falcon-like grasping arm in stage 1: (5) in, To align the arm swing with the desired angle in a falcon-like hunting motion, it is set to... , The positive error boundary should be determined based on the actual situation and the solver performance. This is the angle relaxation factor. The process takes into account the completed target grasping, and by maintaining a constant speed during arm swinging, the system adjusts the robotic arm to a state more suitable for carrying the target during dynamic movement, thereby improving the efficiency of target-carrying flight.

[0015] S33, Phase 2 Planning Constraints for Arm-Mounted Unmanned Aerial Vehicles with Falcon-Inspired Return-to-Home Swing Arms (S33, Phase 2) This phase primarily considers how a falcon, after seizing its target, adjusts its arm swing motion based on the combined kinetic energy of itself and the prey during its return to high altitude, ensuring the prey remains in the air. The planning constraints for the system during this process are: (6) in, , , The positive error boundary should be determined based on the actual situation and the solver performance. , For the target velocity, the constraint also includes the second formula of formula (3), that is The angle design in this section is based on the angle in Phase 1, and the angle is optimized only through energy cost without specific constraints on the angle, aiming to ensure optimal energy during the dynamic process.

[0016] Step 4: Dynamic Planning Constraints for Arm-Mounted Unmanned Aerial Vehicle Systems Due to their inherent characteristics, falcons are flexible in their hunting process and have a wide range of movement, while drones with arms have limited space for maneuver due to mechanical constraints. In order to ensure that the planned instructions conform to the actual system characteristics, this invention considers adding the dynamic constraints of the actual system on the basis of the above optimization problem and motion constraints, so as to better adapt to the actual motion execution of drones with arms.

[0017] S41. Kinematic Constraint Design of Unmanned Aerial Vehicle Systems Quadcopter drones utilize four motors to provide lift for flight, and adjust this lift for flight control. Considering that quadcopter drones typically have sophisticated autopilots capable of receiving desired position, velocity, or acceleration and automatically adjusting lift for stable flight, this section primarily focuses on limiting the velocity and acceleration of desired commands to ensure the drone can execute them correctly. Therefore, the design constraints are as follows: (7) in, This is the upper limit of speed. The upper limit of acceleration is determined based on the actual situation. This constraint ensures that the generated instructions can meet the requirements of actual use and guarantees the feasibility of the algorithm.

[0018] S42, Kinematic Constraint Design of Robotic Arm System The robotic arm of the UAV studied in this invention is mounted below the center of gravity of the UAV. In order to ensure its normal operation and equipment safety, the executable angle of the robotic arm joints is constrained: (8) in, The angle of the first joint of the robotic arm. The second joint angle is defined as follows: the first joint of the robotic arm is fixed to the drone, and the rotation axis is the y-axis of the drone's body coordinate system. The first link is defined as 0° when it is horizontal, and downward swing rotation is the negative direction. The second joint of the robotic arm is connected to the end of the first link, and the rotation axis is the same as the first joint. The second link is defined as 0° when it is in the same straight line as the first link, and downward swing rotation is the negative direction. This angle requires the first link of the robotic arm to swing below the drone, while the second link can move in both positive and negative directions, ensuring the freedom and safety of the robotic arm's movement.

[0019] Meanwhile, to ensure the servo motor's operability, the angular velocity and angular acceleration of the robotic arm were limited: (9) in, For the reason and The vector formed This represents the maximum angular velocity. For the reason and The vector formed This represents the maximum angular acceleration; the relevant limitations are determined by the inherent performance of the robotic arm in actual use.

[0020] Step 5: Solve the optimization problem of near-ground target grasping by the UAV with arm and generate the trajectory. Based on the optimization problem designed to mimic the grasping action of a falcon hunting arm, the final planning result needs to be obtained by solving the optimization problem. In this process, the solver is required to perform algorithmic solutions for the optimization objective in formula (2) and the optimization constraints in formulas (3) to (9). When there are contradictions in the optimization solution, the relaxation factor can be adjusted to ensure the normal execution of the solution. The optimizer will integrate the results of the entire process and generate the optimized state variables by adjusting the control variables. .

[0021] Step Six: Output the grasping trajectory based on the hunting arm swing behavior of a falcon. Based on optimized state variables, the desired position and velocity of the drone with arm at each moment, as well as the desired angle and angular velocity of the robotic arm, are provided to ensure that the drone's own controller can execute the desired commands, ultimately realizing the grasping trajectory based on the hunting arm swing behavior of a falcon.

[0022] This invention proposes a near-ground target grasping planning method for a drone with an arm based on the arm-swinging behavior of a falcon hunting. Its advantages and functions are as follows: 1) It establishes a falcon hunting analysis and a drone system model, and constructs a mapping relationship between the two, providing theoretical support and interpretability for optimizing the grasping trajectory of the drone with an arm; 2) It transforms the trajectory planning problem into a multi-stage energy optimization problem by combining the energy principle of falcon hunting and the energy calculation method of the drone with an arm, and ensures solvability by designing relaxation factors; 3) It designs motion planning constraints for the drone with an arm, mimicking the arm-swinging actions at different stages of falcon hunting, thereby ensuring that the grasping process is smoother, more efficient, and safer. Attached Figure Description

[0023] Figure 1 This is a stage division diagram for an arm-mounted drone based on the hunting arm swinging behavior of a falcon.

[0024] Figure 2 A flowchart illustrating the planning process for near-ground target capture by an arm-mounted UAV based on the hunting arm-swinging behavior of a falcon.

[0025] Figure 3 The entire process of planning and executing three-dimensional trajectory results for near-ground target capture by a drone with an arm.

[0026] Figure 4 To capture the expected position curve of a planned UAV for near-ground target acquisition by a UAV with an arm.

[0027] Figure 5 The desired speed curve of the drone is planned for near-ground target capture by the drone with arm.

[0028] Figure 6 The desired angle curve of the robotic arm is planned for near-ground target capture by a drone with an arm.

[0029] Figure 7 The desired angular velocity curve of the robotic arm is planned for near-ground target capture by a drone with an arm. Detailed Implementation

[0030] The effectiveness of the proposed near-ground target grasping planning method for a drone with an arm based on the hunting arm behavior of a falcon is verified through a specific example below. In this example, the initial position of the drone with the arm is set to (0,0,0), the initial angle of the robotic arm is (-pi / 2,0), there is a velocity of 10 m / s in the positive x-direction, the angular velocity of the robotic arm is 0, there is no acceleration, and the coordinates of the near-ground target to be grasped are set to (100,0,-100). The simulation environment for this example is configured with an AMD 6800 processor, 2.00 GHz clock speed, 32 GB of memory, and MATLAB 2023a software.

[0031] A near-ground target grasping planning method for a drone with an arm, based on the hunting arm swing behavior of a falcon, is illustrated in the flowchart below. Figure 2 As shown, the specific practical steps for this example are as follows: Step 1: Optimize the near-ground target grasping plan of the UAV with arm. S11. Initialize the arm-mounted drone optimization system status Based on the status of the drone with the boom arm, the initial state of the optimization system is set as follows: The initial control quantity is The initial value of the relaxation factor is 0.001.

[0032] S12, Design an optimization problem for near-ground target capture. Based on formula (2), design the trajectory planning and optimization problem for a drone with an arm to grasp an object, where the total time is set to... Optimize sampling time to The relaxation factor proportionality coefficient is 1000.

[0033] S13, Relaxation Factor Design Based on the equations about the relaxation factor in formulas (3) to (5), the initial value of the relaxation factor is designed to be 0.001. It is required that when the system solves a problem, the feasibility of the solution is ensured by adjusting the relaxation factor.

[0034] Step Two: Falcon Hunting Analysis and Modeling of the Arm-Mounted Unmanned Aerial Vehicle System S21. Establishment and State Update of Motion Model for Armed Unmanned Aerial Vehicle Based on the current state and control variables (drone acceleration and robotic arm angular acceleration), the updated drone speed and robotic arm angular velocity are calculated. The system state is updated using Formula 1 to obtain the end position of the drone with arm, thus realizing the current state update. The updated state is then used to solve the optimization problem.

[0035] S22. Phase Division of Near-Ground Target Acquisition Task Based on Falcon Hunting with Arm-Mounted UAV Following the three stages of a falcon's hunting process—pre-grab, during-grab, and post-grab—the target-grabbing task of the drone with an arm is divided into three parts: Stage 0, Stage 1, and Stage 2, with a completion time set for Stage 0. s, the completion time for Phase 1 is s, the completion time for Phase 2 is Each part has independent constraints, and the switching between different stages includes time conditions and location conditions.

[0036] Step 3: Design planning constraints for mimicking the hunting arm swing behavior of eagles. S31, Stage 0: Planning Constraints for a Falcon-Inspired Swing-Arm Approaching Unmanned Aerial Vehicle with Arm Considering that eagles typically transition from a dive to a smooth lateral grasping state before approaching their prey, and pre-align themselves with the prey through arm swinging, constraints on angle alignment, speed alignment, and position alignment are set. The optimization problem during task execution phase 0 must satisfy formula (4). Error boundary , , , , , , All are set to 0.1.

[0037] S32, Phase 1: Planning Constraints for an Arm-Mounted Unmanned Aerial Vehicle Inspired by a Falcon-Style Grasping Arm Considering that during the capture of prey, the falcon transitions from aiming and grasping to loaded flight through a backward arm swing, speed maintenance and robotic arm angle backward swing constraints are set, requiring the optimization problem to satisfy formula (5) during the execution of task phase 1. All error boundary variables are set to 0.1. S33, Phase 2: Planning Constraints for a Falcon-Inspired Return-to-Home Swing Arm-Based Unmanned Aerial Vehicle (UAV) Considering that after a falcon catches its target, it will adjust its arm swinging motion according to the combined kinetic energy of itself and the prey it is carrying during its return to high altitude, speed alignment and position alignment constraints are set, requiring the optimization problem to satisfy formula (6) during the execution of task phase 2. All error boundary variables are set to 0.1. Step 4: Dynamic Planning Constraints for Arm-Mounted Unmanned Aerial Vehicle Systems S41. Kinematic Constraint Design of Unmanned Aerial Vehicle Systems To ensure the feasibility of the actual system, the speed and acceleration constraints of the UAV are designed according to the research object as formula (7), where the maximum speed of the UAV is 16 m / s and the maximum acceleration is 5 m / s². 2 The optimization problem must meet this requirement in all three stages.

[0038] S42, Kinematic Constraint Design of Robotic Arm System To ensure the normal operation of the robotic arm and the safety of the equipment, the robotic arm must satisfy formulas (8) and (9) during its movement, where the maximum angular velocity of the robotic arm is 1 rad / s and the maximum angular acceleration is 0.3 rad / s². 2 The optimization problem must meet this requirement in all three stages.

[0039] Step 5: Solve the optimization problem of near-ground target grasping by the UAV with arm and generate the trajectory. To obtain the desired instructions generated by the plan, it is necessary to address the optimization problem designed in Step 1, combining the motion characteristic constraints of the falcon hunting arm and the motion constraints of the drone with the arm designed in Steps 3 and 4. The Casadi solver in MATLAB is used for problem design, constructing the system optimization variables, objective function, and solver type. The maximum number of iterations is set to 1000, the principal convergence tolerance is 1e-4, the acceptable convergence tolerance is 1e-4, and the constraint violation tolerance is 1e-6. The interior-point method is used for the solution.

[0040] Step Six: Output the grasping trajectory based on the hunting arm swing behavior of a falcon. After configuring the solver, the solver can obtain the solution results that meet the current optimization problem and constraints, namely the system optimization variables, including the expected position and speed of the drone with arm and the expected position and speed of the robotic arm. If the actual usage requirements are not met, the initial state variables and constraint variables such as relaxation factors are reset, and steps one to six are repeated to obtain a new set of optimization results until the planning instructions corresponding to the optimization results meet the actual requirements.

[0041] Figure 3 This is the 3D trajectory result of the entire process of near-ground target grasping planning for the UAV with arm in the current initial state. Blue represents the body trajectory, green represents the first segment of the robotic arm's end effector trajectory, red represents the second segment of the end effector trajectory, green dots represent the target point to be grasped, pink triangles represent the return point to the target. Black dots represent the arrival at point p1, and black squares represent the arm swing completion point.

[0042] Figure 4 and Figure 5 The desired position and velocity curves of the drone are planned for near-ground target capture by the drone with arm. The triangle represents the state corresponding to point p1 and the circle represents the state corresponding to point p2. From the curve of the z-axis, it can be seen that the trajectory smoothly transitions to the target position, thereby avoiding the impact force in the z-direction during capture and improving capture efficiency and safety.

[0043] Figure 6 and Figure 7The desired angle and angular velocity curves of the robotic arm are planned for near-ground target grasping by a drone with an arm. Triangles represent the state corresponding to point p1, and circles represent the state corresponding to point p2. It can be observed that this process includes alignment in stage 0, swinging in stage 1, and restoring to the default state in stage 2. This demonstrates that this process can model the arm swinging behavior of a falcon hunting, thereby improving the efficiency of grasping tasks.

Claims

1. A near-ground target grasping planning method for an arm-mounted UAV based on the hunting arm swing behavior of a falcon, characterized in that: The steps of this method are as follows: Step 1: Falcon Hunting Analysis and Modeling of Arm-Mounted Unmanned Aerial Vehicle System S11. Analyze the arm-swinging behavior of falcons during hunting; The process of hunting with a hawk arm is divided into three stages: before grasping, during grasping, and after grasping. Based on the characteristics of hawk behavior in each stage, optimization constraints that conform to the motion model of the hawk-armed drone are established. Based on the energy design optimization problem, trajectory generation is achieved. S12. Establish a basic model of the drone with arm; S13. Stage division of the arm-mounted drone grasping task; Step 2: Optimize the near-ground target acquisition planning of the UAV with arm. The guidance planning problem is transformed into a planning optimization problem. The optimization objective is established through energy relationships, while ensuring that the inherent physical constraints of the UAV and the behavioral characteristics of the falcon's arm swing are satisfied before, during and after the grasping process. S21. Design an optimization problem for near-ground target capture; S22, Position Relaxation Factor Design; Step 3: Design planning constraints for mimicking the hunting arm swing behavior of eagles. S31, Phase 0: Planning constraints for a swing-arm-oriented UAV model inspired by a falcon. S32, Phase 1 Planning constraints for a UAV with an arm that mimics the grasping arm of a falcon; S33, Phase 2 Planning constraints for UAVs with arms that mimic the return swing arm of a falcon; Step 4: Dynamic Planning Constraints for Arm-Mounted Unmanned Aerial Vehicle Systems S41. Kinematic constraint design of the UAV system: By limiting the speed and acceleration of the desired command, the UAV can be executed normally. S42. Kinematic constraint design of the robotic arm system, specifically constraining the executable angles of the robotic arm joints; Step 5: Solve the optimization problem of near-ground target grasping by the UAV with arm and generate the trajectory. The final planning result is obtained by solving the optimization problem. The solver is required to solve the optimization objective and the optimization constraints of the falcon hunting arm. When there are contradictions in the optimization solution, the relaxation factor is adjusted to ensure the normal execution of the solution. Step Six: Output the grasping trajectory based on the hunting arm swing behavior of a falcon. Based on optimized state variables, the desired position and velocity of the drone with arm at each moment, as well as the desired angle and angular velocity of the robotic arm, are provided to ensure that the drone's own controller can execute the desired commands, ultimately realizing the grasping trajectory based on the hunting arm swing behavior of a falcon.

2. The near-ground target grasping planning method for an arm-mounted UAV based on the hunting arm swing behavior of a falcon, as described in claim 1, is characterized in that: The first step of the arm-mounted UAV grasping task phase is divided into three phases: the arm-mounted UAV approach phase, the target grasping phase, and the grasping and carrying flight phase, referred to as phase 0, phase 1, and phase 2.

3. The near-ground target grasping planning method for an arm-mounted UAV based on the hunting arm swing behavior of a falcon, as described in claim 1, is characterized in that: The specific process of step two is as follows: S21. Design an optimization problem for near-ground target capture. The design optimization problem for a drone with an arm for grasping is as follows: (1) in, The cost function value, Total execution time Unmanned aerial vehicle system motion vector, , The angular velocity of the robotic arm is given by [value], and the length of the first segment of the robotic arm connected to the drone is [value]. The length of the second robotic arm is l 2 , , , , All are relaxation factors. The relaxation factor is the scaling factor, which is used to constrain the reach of the winged drone to the target. S22, Location Relaxation Factor Design In the specific optimization of the arrival design, a relaxation factor was designed to reduce the constraints of the arrival conditions. Firstly, the target position to be captured upon completion of stage 0 was set as... When Phase 2 is completed, the end position of the robotic arm of the drone will be... Based on this, the relaxation factor is defined as: (3) in, For the position vector of the armed UAV, For stage 0, the position of the drone with arm is captured at the moment of execution. To determine the end position of the drone with arm after the capture and return time in Phase 2, a relaxation factor is set to relax the constraints on the arrival position.

4. The method for controlling drone swarms based on eagle-dove game theory according to claim 1, characterized in that: The specific process of step three is as follows: S31, Stage 0: Planning Constraints for Arm-Mounted Unmanned Aerial Vehicles Approaching a Falcon-like Swing Arm (S31, Stage 0) At the end of phase 0, it is required that the end effector velocity of the drone with arm has no vertical velocity, and the end effector of the robotic arm should be oriented horizontally. The motion constraints are as follows: (4) in, To align the arm swing with the desired angle in a falcon-like hunting motion, it is set to... , , , The positive error boundary should be determined based on the actual situation and the solver performance. , Set the target speed to 0. For the angle relaxation factor, The swing angle of the robotic arm at the grasping moment in phase 0. , , For the x, y, and z velocities of the robotic arm's end effector at the grasping moment in Phase 0, the drone with the arm must maintain the same velocity as the target at the end of Phase 0 and align itself according to the arm swing characteristics of a falcon's hunting preparation phase. Furthermore, this phase constraint also includes the formula... ; S32, Phase 1 Planning Constraints for Arm-Modified UAV with Falcon-Inspired Grasping Arm The constraints for the falcon-like grasping arm designed for Phase 1 are as follows: (5) in, To align the arm swing with the desired angle in a falcon-like hunting motion, it is set to... , The positive error boundary should be determined based on the actual situation and the solver performance. For the angle relaxation factor, The angle of the robotic arm after the completion of Phase 1 swing arm. This refers to the end-effector velocity vector during the grabbing process in phase 0. The end velocity vector after the swing arm completes the swing arm in stage 1. Considering that the target grasping has been completed, the swing arm is performed while keeping the velocity constant, so that the system can adjust the robotic arm to a state that is more suitable for carrying the flight during dynamic movement, thereby improving the flight efficiency of carrying the target. S33, Phase 2 Planning Constraints for Arm-Mounted Unmanned Aerial Vehicles with Falcon-Inspired Return-to-Home Swing Arms (S33, Phase 2) The planning constraints for the system design in this phase 2 are: (6) in, , , The positive error boundary should be determined based on the actual situation and the solver performance. , For the target velocity, the constraint also includes the formula ; , , The x, y, and z velocities of the robotic arm's end effector at the moment of return after grasping are determined. The angle design is adjusted based on the angle in Phase 1, and the angle is optimized only through energy cost. No specific constraints are imposed on the angle to ensure optimal energy during the dynamic process.

Images